Analyses based on annual supply and demand of gold appear on a daily basis, whether posted to gold web sites or in the financial media, many of them by the most respected analysts of gold mining shares. These articles typically show an imbalance between supply and demand, suggesting that there is a gold supply deficit. From there, the conclusion follows that a much higher gold price is required in order to bring supply and demand into balance.

There is no gold supply deficit. Even if there were, to cite Dick Cheney, "deficits don't matter". The dollar price of gold is formed through the balancing of total gold supply and demand against total dollar supply and demand. The incremental supply and demand during any one-year period is irrelevant to the price. The illusion of a deficit comes about from an incorrect interpretation of supply and demand figures: annual amounts rather than totals are compared.

On the supply side, the annual production of gold has almost nothing to do with its price. Neither does decades of under-investment in gold exploration, the lack of new discoveries of gold deposits, miners' cash cost per ounce, nor environmental delays in permitting new mines. The output of the gold mining industry has very little impact on the gold price.

On the demand side, the "annual demand" for gold -- as it is computed in the models showing a deficit -- is a misleading figure. The comparison of annual amounts is relevant for a commodity that is consumed but not one that is held as is gold. For an asset that is held, the annual demand has no business being compared against annual supply, for the comparison tells us nothing about the price.

Whose Deficit?

While I could cite hundreds of examples if I had been collecting them over the years, in the interest of space, I will cite only two to make my point. These two examples were selected not to single out the particular writers, as there were many others that could have been chosen, but because they happened to pass in front of me recently.

First, this article on a mining site:

GOLD supply shortages were possible in the long-term, according to recent research produced by Canadian research house, Metals Economics Group (MEG). It said in a press statement that recently discovered deposits of more than 2.5 million ounces, enough to attract the interest of major gold producers, were not adequate to replace their production.

And this piece from a financial news site:

...JP Morgan believes the gold market outlook continues to improve. Demand continues to strengthen (even if only for one-off events such as the establishment of gold Exchange Traded Funds or ETFs), but this stronger demand is not being met by higher supply thanks to declining production from South Africa in particular. This means central bank selling is required to meet the shortfall, but the quantity of this selling is limited under agreements in place between the banks.

The Case for a Deficit

In order to understand why there is no deficit, I will explain why some people think that there is one.¹ The problem with the supply deficit theory is in the interpretation of the numbers, and not the numbers themselves. Because the exact numbers don't matter so much, I will use the gold supply and demand figures from a prominent industry source, the World Gold Council, without attempting to verify them. Values for the last three years are found in their supply and demand spread sheet.² Even if slightly different numbers were used, the point that I am going to make would not change.

Table 1, below, is based on the WGC figures for the last two years. Note that they do not show much of a deficit for 2004 and a slight surplus for 2005. The WGC, as far as I know has not promoted the supply deficit argument. However, I am citing their figures because they use the annual supply and demand methodology, the same methodology that is used by analysts who think that there is a deficit.

Table 1: WGC Annual Figures

A report from the UK branch of the French bank Cheuvreux caused considerable discussion when it was released last year. This report, using the same flawed methodology, showed much larger supply deficits on an annual basis. It is worthwhile to understand the discrepancy between these two reports. On pages 26 and 27 of the report, the information used to construct Table 2, below, appears. While the WGC shows a surplus for both 2004 and 2005 on the bottom line a report from the Cheuvreux report, while using essentially the same numbers as the WGC, shows estimated supply shortfalls of hundreds of tons annually.³

Table 2: Cheuvreux Annual Supply and Demand

The difference between the two reports using the same raw data are substantial and must be explained. The main source of the WGC definition of supply value includes official sector sales while the Cheuvreux definition of supply does not. In the Cheuvreux report, the net supply minus demand (which they call supply shortfall) is greater than the net of supply minus demand in the WGC report by an amount approximately equal to the size of official sector sales. Because the official sector sales are a fairly large number, the Cheuvreux value for net of supply and demand is a negative number in both 2004 and 2005.

Cheuvreux shows the official sector sales in a separate row appearing in their table after Supply Shortfall. By removing official sector sales from the supply, this format implies that official sector sales were necessary in order to fill a deficit between the other components of supply and the demand. While official sector sales offered "at market" probably do affect the gold price, this impact is exaggerated by offsetting official sales against annual figures rather than totals.

Deficits Don't Matter

Let's look at how the WGC and the Cheuvreux arrive at a deficit.

In the WGC report, a footnote states (with some caveats) that the Balance term is partly due to residual error (presumably errors in measurement); and that the remaining Balance is the "implied value of net (dis) investment" ("includes institutional investment other than ETFs and similar stock movements"). In the WGC report, a negative Balance (deficit) would occur in any year where there are net private (non-official) sales.⁴

The Cheuvreux report starts from the position of the WGC report, however, Cheuvreux does not include the additional differential due to the omission of official sector sales from their definition of supply. Cheuvreux defines a deficit year as any year during which there were net private plus official sector sales.

A word can be defined to mean anything, but is the definition useful? I will argue that to define a deficit year as a year in which there are private sector or official sales is more than a little bit misleading, because it leads to thinking about the gold market as if it were a spot market for a commodity that is consumed rather than held.

For a commodity that is consumed, an annual incremental deficit would imply a higher price in the future because the deficit could only be filled by a drawdown of existing stockpiles, which would eventually become exhausted if the deficits continued. Upon the depletion of stockpiles, the price would have to rise to the point where demand was in balance against only that supply that was produced.

But gold is not that sort of commodity. There is no need at all for supply on an annual basis excluding private sales to come into balance with demand on an annual basis. It is not even true that these must balance over any number of years. The reason for this is that a sale out of someone's stockpile of gold does not reduce the total amount of stockpiled gold. All it does is to shift the gold from the seller's private stockpile to the buyer's private stockpile. A market could remain in a "deficit" of this sort forever without the price ever going up (or going down) as buyers and seller shifted the contents of their stockpiles among themselves.

Stocks and Flows

We can divide economic goods into those for which the entire annual supply is destroyed in the process of consumption, and those for which new supply is hoarded.⁵ Economists call the former "flows" and the latter "stocks".

Analyzing the supply and demand over a short window of time for a flow-type good would tell you a lot about where the price was likely to go. But annual supply and demand for the second type - of which gold is the premier exemplar - tells you almost nothing about its future price movement.

First, consider a good that is consumed, where by "consumed" I mean that the economic value of a unit is destroyed over the course of its productive life. One example is DVD players. The economic value of a player is destroyed as the player wears out. All of the supply that manufacturers produce must be sold. There would be no real reason for Sony to sit on warehouses full of aging players. The price of the players can only fall as they become obsolete, and on top of that, they are costly to store. Sony must sell everything that they produce at whatever price the market will support at the time.

Competition from other manufacturers to sell, and competition among consumers to buy Sony's players, or other goods entirely, ensures that the price at which the players are sold will be whatever price clears the market between all buyers and sellers on a very short time scale. In micro-economic jargon, most final goods have a vertical supply curve once they arrive at the market. The same would be true of any perishable good, most manufactured goods, and commodities that can only be stored for a short time, such as beef or eggs.

But for most known commodities, the aboveground supply is relatively small compared to the quantity that is permanently used up every year. Most of what is mined, drilled, grown, or raised on a farmed is consumed soon after it is produced. In some cases, large stockpiles of a particular metal - e.g. silver -- have been accumulated and in other cases accumulated stockpiles have sold off (silver again). But absent a large stockpile the market price of these goods is pretty close to the level that balances the recent supply and current demand.

When it comes to a stock, total (not annual) supply and demand determine the price of each unit. Consider the following example concerning equity shares of a corporation. Suppose that an equity analyst appeared on CNBC stating that the price of a common share in company XYZ, with 100M shares issued, would rise (or fall) because they were only issuing 1M new shares this year, while the demand for those shares would be 2M. This analyst would be pricing the shares as if they were a stock-type of good. Using this method, a daily volume of 1M shares would be an annual volume of about 250M, which would create a "supply deficit" of 249M shares assuming 1M new shares issued.

It is easy to see the fallacy here. Even if the capital raised from issuing the new shares added no value at all to the corporation,⁶ at worst it would only dilute the value of the existing shares by 1%. A stock with 100M shares outstanding could easily trade 1M shares per day without the price rising or falling as people rearrange their portfolios with some who wish to hold fewer shares selling, and other investors who wish to hold more shares buying.

The True Supply of Gold

To understand the price of gold, the relevant supply is the total supply, not the new supply coming to market during the last year (or week or month). The supply of gold consists of all of the supply that exists. The relevant demand is the total demand, not the new demand coming to market during any year.

For gold, there is always a large stockpile, and it never gets smaller. The vast majority of all gold mined throughout human history still exists and is held either in bars, coins, or jewelry. According to the WGC, this quantity was around 155,500 tonnes at the end of 2005. Almost no gold is used up (in the sense of being destroyed or becoming permanently unusable) ever. In most cases when a buyer purchases gold, it moves from the seller's hoard to the buyer's hold.

The World Gold Council estimates that 52% of gold is held as jewelry. James Turk subdivides jewelry holdings into low carat and high carat. The former is purchased mainly for the gold value, as an alternative to buying bars and coins. The latter is purchased mostly for fashion. According to Turk's estimate (which was published in 1996), monetary jewelry at that time accounted for about 60% of jewelry with fashion jewelry accounting for the remaining 40%. However, even when made into jewelry, the gold is not destroyed and can come back into the market as scrap. The WGC figures show significant recovery from scrap.

The reason that total supply and not annual supply matters is that the gold market is not segregated into two markets. There is not one gold market for the current year and another gold market for aboveground gold that was mined in previous years. The gold market is a single market in which all sources of supply are indistinguishable. Every existing ounce of gold competes for sale with every newly mined ounce. A buyer of gold doesn't care whether he is buying recently mined gold or gold that was held in bars for 100 years, or the product of melted jewelry.

Every ounce of gold that is held by someone is potentially for sale at some price. While not every ounce of gold in private hands is for sale at the current market price, any ounce of gold could potentially come to market. A lot of gold is held in small stockpiles among widely dispersed owners. Some is for sale just above the current spot price, some only at much higher prices. The varying levels of prices at which different units of goods held in a stock are offered for sale is what makes the supply curve upward-sloping rather than vertical as is the case in consumption goods.

Is it true that a lot of gold is not for sale at all, so it should not be counted as part of the supply? In short, no. gold is held as a store of value over time. The point of holding a store of value is not to hold it forever and then have it cremated along with your corpse. A person will only store value over time because they anticipate the need for the value some time in the future. Anyone who anticipated having no needs in the future would not need to store value over time. And the stored value is only stored for a finite period of time until the person holding it becomes aware of something that they need more than what they have stored. That would be the time to sell.

Note also that every new ounce of gold that is mined does not need to be sold at the current market price. Unlike most manufactured goods, gold mining companies do not necessarily have a vertical supply curve for their product because it does not spoil or become obsolete. While many mining companies do sell all of their supply at spot soon after they have mined it, some mining companies sell their supply at a pre-determined price that in some cases was fixed years in advance through hedging contracts. And other mining companies choose to hold mined supply in reserve with the anticipation of selling it later, at a higher price. Goldcorp has done this in the past, at one point accumulating more vault gold than the central banks of a large number of small nations.

The Demand for Gold

It is easy enough to see that the supply of gold is the total supply. But what is the demand? It turns out that the demand is equal to the supply. To understand this, we introduce the concept of reservation demand. Most people are familiar with exchange demand. Exchange demand is expressed by giving up something in an exchange in order to for the thing demanded. Reservation demand is a demand that is expressed by holding onto something that you own.

People who hold gold are demanding it by holding it off the market. As Austrian economist Murray Rothbard explains,

At any point on the market, suppliers are engaged in offering some of their stock of the good and withholding their offer of the remainder. ... This withholding is caused by one of the factors mentioned above as possible costs of the exchange: either the direct use of the good (say the horse) has greater utility than the receipt of the fish in direct use; or else the horse could be exchanged for some other good; or, finally, the seller expects the final price to be higher, so that he can profitably delay the sale. The amount that sellers will withhold on the market is termed their reservation demand. This is not, like the demand studied above, a demand for a good in exchange; this is a demand to hold stock. Thus, the concept of a "demand to hold a stock of goods" will always include both demand-factors; it will include the demand for the good in exchange by nonpossessors, plus the demand to hold the stock by the possessors. The demand for the good in exchange is also a demand to hold, since, regardless of what the buyer intends to do with the good in the future, he must hold the good from the time it comes into his ownership and possession by means of exchange. We therefore arrive at the concept of a "total demand to hold" for a good, differing from the previous concept of exchange-demand, although including the latter in addition to the reservation demand by the sellers.

The Total Picture

Now that we have covered the total supply and total demand, the proper rendering of the supply and demand situation would look something like Table 3, though the numbers are not exact. Note that when all sources of supply and demand are counted, there is no deficit. Total supply and total demand must always equal because every transaction has a seller and a buyer. Over time, there is a gradual accumulation of the stock of gold and a possible shifting between investment holdings (bar, coin, ETF) and jewelry.

Table 3: Total Supply and Demand

The price of gold is determined as is the price of any stock: by total supply and total demand. The price is that price which balances total supply against total demand, including reservation demand. The price of gold, in terms of dollars, or other fiat money, balances supply of all gold offered for sale at a range of prices in dollars with demand for gold - including both demand to exchange dollars for gold and the reservation demand for dollars and for gold.

Looking at supply and demand over a single year tells us nothing because the annual supply and demand are only about 2-3% of the total supply and demand, while the price of gold depends mostly on the other 98%.

Suppose that during a particular year, there are net sales from stockpiles. This tells us nothing about what the price of gold will do, because when gold is sold, it goes from the seller's private stockpile into the buyer's private stockpile. There is no limit on the number of consecutive years in which sellers of gold can sell out of their stockpiles as long as there are buyers who add to their stockpiles that same year. This type of trade in gold could go on forever without the price changing because individuals' needs change all the time. During a given year, there will always be some people who have an increasing need of a store of value and others having a decreasing need. The former become buyers, the latter, sellers.

Annual changes to supply and demand do not influence the price much, if at all, because annual changes are small compared to the total. Around 98% of supply during any year was previously mined. And around 98% of demand is reservation demand, while only around 2% of demand is exchange demand for mined gold.

Newly mined gold does have some effect on the gold price, but only insofar as it dilutes the total supply of gold by a small amount. As previously discussed, mine supply dilutes existing supply by about 2% annually. If the supply of gold were diluted by 2% each year, and existing holders wanted to hold the same amount of gold in their portfolios measured in purchasing power terms, then existing holders would rebalance their portfolios adding about 2% to their positions and the price of gold would have to fall by about 2%.

If gold mining were to increase by 50% from the current year to the next year, would the price of gold collapse? Not at all. After a 50% increase, the proportion of new mine supply out of total supply would only rise from 2% to 3%. Gold demand would not need to increase by 50% in terms of ounces to absorb this supply, only by about 0.98% (1.03/1.02 - 1.0).

But even this overstates the influence of newly mined gold. It is probably more relevant to measure demand in dollar terms rather than in ounces. In this example, if the price of gold in dollars declined by about 0.97% (1/0.98) while gold demand in dollars remained constant, the demand in ounces would increase by just enough to balance the new supply. With a total demand, properly counting reservation demand, absorbing the newly mined gold into the market doesn't appear nearly so difficult.

Another way of making the same point is to suppose that gold mining stopped entirely. Investment demand by new investors in gold would have to be met by an equal amount of disinvestment by existing holders. In that case, then every buyer would have to buy gold from a private or official sector seller. If an annual deficit year is defined as one in which there are net private sector sales, the market would be in deficit every single year. No matter how high the price moved, the market would still be in deficit. There is no price of gold that would cure the deficit because of the way that the deficit is defined. But the price would not necessarily go up under these conditions because any sales out of a seller's stockpile are exactly offset by additions to a buyer's stockpile. All that happens in a market like this is that stockpiles change ownership from owners who value them less at that time to owners who value them more. No general statement about the price can be made; however, during periods of the classical gold standard, the purchasing power of gold tended to rise by a few percent per year.

Silver is Not Gold

When it comes to silver, deficits do matter. Here I cite the works of silver analysts David Morgan, Ted Butler, and Charles Savoie. For most of the past few thousand years, annual mine supply was in equal or in surplus over annual consumption, and stockpiles were accumulated year after year, at one point reaching around 6 billion ounces. Over the last forty years, stockpiles have been drawn down to nearly zero. At the present time, all of the silver consumed during any given year is silver that came out of the ground that year, with a decreasing contribution from stockpiles.

With silver, it does make sense to look at net private sales from stockpiles as filling a deficit between supply and demand. Why is this true for silver and not for gold? For the most part, demand for silver is consumption demand, and most consumption demand is destructive, meaning that the silver ends up in a form where it cannot easily be reclaimed and brought back into the market.

Therefore, the deficit of mine supply relative to destructive demand implies a necessary sale out of stockpiles. These finite stockpiles cannot continue to supply the world with 100Moz of silver for consumption annually. At some point, they must be exhausted and higher prices will then be required in order to bring supply and demand into balance on an annual basis.

To the extent that the silver held for investment purposes, then everything I have said about gold applies to silver. The same would be true for photographic demand for silver because most silver used in photography is reclaimed. Silver is partly held for investment purposes and partly consumed, so its price behavior will result from a combination of the two models.

Conclusion

Does the non-existence of a supply shortage theory make the case for gold weaker? I say no. At the beginning of this article, I stated that there is a bullish case for gold. If not the mythical shortage of mined supply, then what is it?

Analysts including Frank Veneroso, Reginald Howe, Robert Landis, John Embry, and others affiliated with the GATA organization have shown in a series of research reports published over the last five years that central banks have created what amounts to a large naked short position in the gold market using paper derivatives. The accumulation of shorts without any offsetting longs has been a negative for the gold price, especially during the late 90s.

But the ultimate bullish case for gold is none other than the bearish case for fiat money, the dollar, and central banking. Gold is money and while central banks have the ability to debase fiat money up to a point, they are in the end limited by the acceptability of their paper as money. The end game of the paper monetary system is collapse and its replacement by the natural monetary order of gold.

Notes:
1 To be fair, I believe that the gold analyst community is divided on this issue. There are many analysts who use the correct method of comparing total supply and total demand. What prompted me to write this piece is the disconnect that I see between the two groups. I have listened to interviews between analysts on opposite sides of this issue where they each went away thinking that the other agreed with them.
2 This site may require free registration to view some items.
3 The Cheuvreux report cites the World Gold Council as the source of the data but the annual numbers differ slightly. I am not sure why but the differences are small so it doesn't affect the argument.
4 I should point out that the WGC does not use the term "deficit", though many writers using these figures, or figures like this, do use that term.
5 There are in-between cases, like houses, which are used up over a period of decades. But for analytical convenience, we posit these two ideal types in order to understand the different pricing behavior of each one.
6 Actually, this happens all the time when corporations issue new shares upon the exercise of employee stock options.